Forum: Ruby-core [Ruby-Bug#3780][Open] RDoc::Parser.binary? broken for some utf8 files longer than 1024 bytes

Bug #3780: RDoc::Parser.binary? broken for some utf8 files longer than 
1024 bytes
http://redmine.ruby-lang.org/issues/show/3780

Author: Stephen Bannasch
Status: Open, Priority: Normal
Category: core
ruby -v: ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-darwin10.4.0]

RDoc truncates files at 1024 bytes when checking if the file is binary. 
This will invalidate the file encoding if the file is truncated in the 
middle of a utf8 char and cause RDoc to exit.

I found this problem when running rdoc on the ruby 1.9.2 source.

  $ ruby -v
  ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-darwin10.4.0]
  $ rdoc --version
  rdoc 2.5.11

More description of the bug and a patch with a failing test is on this 
issue in RubyForge rdoc issue tracker.

http://rubyforge.org/tracker/index.php?func=detail&aid=28525&group_id=627&atid=2472

The same issue appears to be in the 1_9 source, see: 
http://github.com/ruby/ruby/blob/trunk/lib/rdoc/parser.rb#L70

I find it confusing knowing where to create an RDoc issue: RubyForge or 
here -- so I've created an issue in both places.

This gist: http://gist.github.com/561350 (possible_fix.rb) shows how I 
changed RDoc::Parser.binary?  locally --  but I don't think it is 
correct to classify all utf8 files which are invalid when truncated at 
1024 bytes as binary.

That same gist (show_parsing_error.rb) also shows another strategy for 
solving the invalid encoding issue but there are probably better ways to 
determine if a file is binary.

Issue #3780 has been updated by Eric Hodel.


RDoc 2.5.11 is newer than the version of RDoc than ships with Ruby 
1.9.2.

RDoc 2.5.8 ships with Ruby 1.9.2.

Can you confirm that this bug exists in the default RDoc that ships with 
1.9.2?
----------------------------------------
http://redmine.ruby-lang.org/issues/show/3780

Issue #3780 has been updated by Stephen Bannasch.


Interesting ... the problem does not occur when running rdoc included in 
ruby built with the v1_9_2_0 tag. I had thought it would -- but the 
RDoc::Parser.binary? method  I reference above which I believe causes 
the problem: 
http://github.com/ruby/ruby/blob/trunk/lib/rdoc/parser.rb#L70  is from 
trunk -- appears to be identical??

$ ./bin/ruby --version
ruby 1.9.2p0 (2010-08-18 revision 29036) [x86_64-darwin10.4.0]

$ ./bin/rdoc --version
rdoc 2.5.8

$ rm -rf ~/Desktop/rdoc; ./bin/rdoc -o ~/Desktop/rdoc 
~/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/
Parsing sources...
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/irb/inspector.rb:36:36: 
Couldn't find INSPECTORS. Assuming it's a module
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/singleton.rb:238:11: 
Couldn't find Yup. Assuming it's a module
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/tk/font.rb:41:27: 
Couldn't find SYSTEM_FONT_NAMES. Assuming it's a module
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/tk.rb:67:30: 
Couldn't find Tk_CMDTBL. Assuming it's a module
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/tk.rb:72:31: 
Couldn't find Tk_WINDOWS. Assuming it's a module
100% [877/877] 
/Users/stephen/.rvm/rubies/ruby-1.9.2-p0/lib/ruby/1.9.1/yaml.rb

Generating Darkfish...

Files:       877
Classes:    1647 ( 1138 undocumented)
Constants:  1894 ( 1630 undocumented)
Modules:     444 (  314 undocumented)
Methods:   12982 ( 9305 undocumented)
 26.99% documented

Elapsed: 285.4s

----------------------------------------
http://redmine.ruby-lang.org/issues/show/3780

Issue #3780 has been updated by Eric Hodel.

Category set to lib
Status changed from Open to Assigned
Priority changed from Normal to Low
Target version set to 1.9.3

Ok, thanks for the confirmation of where the problem occurs.

I've been adding proper encoding support to RDoc and it reveals that the 
current implementation is naive.

The next release should work properly on 1.9.
----------------------------------------
http://redmine.ruby-lang.org/issues/show/3780

Issue #3780 has been updated by Eric Hodel.

Status changed from Assigned to Closed

Fixed by import of RDoc 3.5